You are on page 1of 15

Fluid Phase Equilibria 194–197 (2002) 755–769

Application of density functional theory for predicting the surface


tension of pure polar and associating fluids
Jiu-Fang Lu∗ , Dong Fu, Jin-Chen Liu, Yi-Gui Li
Department of Chemical Engineering, Tsinghua University, Beijing 100084, PR China
Received 10 March 2001; accepted 20 August 2001

Abstract
A thermodynamic method has been developed based on the density functional theory (DFT) to predict the
surface tension of polar and associating fluids by the authors. The Barker–Henderson (BH) perturbation theory
and statistical associating fluid theory (SAFT) are used to establish the equation of state (EOS). The hard sphere
repulsion, dispersion, chain formation, and dipole–dipole or association interactions are taken into account. The
parameters m, σ and ε/k for non-associating polar fluids and m, σ , ε/k, εAB /k and κ AB for associating fluids
are correlated by simultaneously fitting the saturated vapor pressure and the liquid density data with the EOS. The
surface region of a pure liquid is divided into many extreme thin layers. The chemical potential in every layer of the
surface leads to a constant by optimizing the surface thickness. The density profile is obtained from the optimized
surface thickness and the hyperbolic tangent function obtained from molecular simulation. By use of the obtained
density profile and the regressed parameters in EOS, the surface tensions for four pure non-associating polar fluids
and 11 associating fluids in wide temperature range are predicted satisfactorily. © 2002 Elsevier Science B.V. All
rights reserved.
Keywords: Surface tension; Density functional theory; SAFT; Density profile; Equation of state; Polar fluid; Associating fluid

1. Introduction

Interfacial behavior plays an important role in our daily life and chemical engineering. Surface tension
is the characteristic property of fluids. It is important for the development, design and simulation of
many industrial processes. In recent years, more and more studies have been contributed to this field. A
growing number of surface tension data for vapor–liquid surfaces is available because of the development
of new experimental techniques and many new theoretical methods have also been established to solve
the interfacial phenomena.
The traditional Gibbs method is widely used in this field. It is considered as a simple and useful way
to establish the empirical and semi-empirical equations to correlate or predict the surface tension [1–3].

Corresponding author. Tel.: +86-106278-4540; fax: +86-106277-0304.
E-mail address: jflu@ht.rol.cn.net (J.-F. Lu).

0378-3812/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 3 8 1 2 ( 0 1 ) 0 0 6 9 2 - 6
756 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

The perturbation theory is also used for surface tension [4–6]. The parameters in the perturbation theory
have obvious physical meaning and the method is good in prediction. However, the surface is treated as
a two-dimensional plane or a layer with thickness of one molecule and the density gradient in the surface
region are neglected in both methods mentioned above.
The integral equation theory, the density gradient theory and the density functional theory (DFT) can
be used to the heterogeneous fluids. Among those theories, DFT is the most popular approach. The
density gradient, the thickness and the heterogeneity in the surface region are taken into account in this
theory. It has been developed very rapidly in solving the interfacial phenomena [7]. Winkelmann and
coworkers [8,10], and Winkelmann [9] did the research on DFT to model the surface tension of fluids.
Teixeira and Telo da Gama [11] used the DFT and the local density approximation (LDA) to calculate
the surface tensions and density profiles for polar fluids. The dipole–dipole interaction has been taken
into account in their study. Tarazona et al. [12] and Peterson et al. [13] used the DFT and the smoothed
density approximation (SDA) to the equilibrium of surface for polar fluids and compared their results
with LDA. Cai et al. [14] used this theory to confined chain like molecules.
In this paper, a theoretical method based on the DFT is built to predict the surface tensions for polar fluids.
An equation of state (EOS) based on Barker–Henderson (BH) perturbation theory [15] and statistical
associating fluid theory (SAFT) [16,17] is built to regress the parameters. By use of the DFT and the
same set of parameters from EOS, the surface thickness and the density profile are calculated and the
surface tension of pure polar fluids and associating fluids are predicted.

2. Theories

2.1. Density functional theory

The DFT is based on the grand potential functional Ω[ρ(rr )], which can be expressed as follows:

Ω[ρ(rr )] = F [ρ(rr )] + ρ(rr )[Eext (rr ) − µ] drr (1)

where Eext (rr ) is the external potential, µ the chemical potential, drr the infinitesimal, F [ρ(rr )] the
Helmholtz free energy function, ρ(rr ) the number density of the molecules, which is a function of r .
The LDA is applied as follows:

F [ρ(rr )] = [f [ρ(rr )]] drr (2)

where f stands for the Helmholtz free energy density.


For a free surface, no external potential exists, hence E ext (rr ) = 0. Eq. (1) then can be expressed as
follows:

Ω[ρ(rr )] = [f [ρ(rr )] − ρ(rr )µ] drr (3)

When Eq. (3) is minimized, we have {δΩ[ρ(rr )]} /δρ(rr )=0 and the chemical potential µ= {δf [ρ0 (rr )]} /
δρ0 (rr ). The grand potential functional dΩ at constant temperature can be expressed as follows:
dΩ = γ dA − p dV (4)
where p is the normal pressure tensor, which is equal to the vapor pressure.
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 757

Fig. 1. A selected layer in the surface.

The surface of liquid is divided into many extreme thin layers as shown in Fig. 1. The surface tension
in the layer i can be calculated as follows:
dΩ[ρ0 (rr )] + p dV dΩ[ρ0 (z)] + p dV
γi = = zi (5)
dA dV
where ρ0 (rr ) is the equilibrium number density of molecules and A the surface area.
The total surface tension in the surface is summed as follows:
 ∞   n  
dΩ[ρ0 (z)] Ω[ρ0 (z)]
γ = + p dz ≈ + p zi (6)
−∞ dV i=1
V

From Eq. (3), we have


dΩ[ρ0 (z)]
= f [ρ0 (z)] − ρ0 (z)µ
dV
Hence, the surface tension, γ can be expressed as follows:
 ∞
γ = [f [ρ0 (z)] − ρ0 (z)µ + p] dz (7)
−∞

where f [ρ0 (rr )] = F [ρ0 (rr )]/V .


From thermodynamics for pure fluids, we know that
µ F
= +Z (8)
NkT NkT
p[ρ0 (z)] = Zρ0 (z)kT (9)
where p[ρ0 (z)] is the tangential pressure tensor, Z the compressibility factor, N the number of molecules,
k the Boltzmann constant, and T the absolute temperature.
Thus, the surface tension can be expressed as follows:
 ∞
γ = [p − p[ρ0 (z)]] dz (10)
−∞

where p[ρ0 (z)] can be calculated from the density in every layer by the EOS.
758 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

2.2. Equation of state

The BH perturbation theory and SAFT are used to build the EOS for pure polar and chain-like associating
liquids, respectively. The Helmholtz free energy for polar fluid can be expressed as follows:
F − F id = F hs + F chain + F LJ + F dd (11)
where Fid is the Helmholtz free energy of an ideal gas with the same density and temperature as the
system, Fhs the free energy of a hard-sphere fluid relative to the ideal gas, Fchain the free energy change
when chains are formed from hard spheres, FLJ the contribution to the free energy of Lennard–Jones
interaction and Fdd the contribution of dipole–dipole interaction. For associating fluid, Fdd is omitted but
the free energy change due to association (Fassoc ) is taken into account.
The Fid in Eq. (11) is calculated as follows:
F id = F 0 + N0 kT ln ρ (12)
where F0 is the Helmholtz free energy in its standard state. The hard-sphere term Fhs can be expressed
by the Carnahan–Starling equation [18] as follows:
F hs 4η − 3η2
=m (13)
NkT (1 − η)2
where η = (π/6)mρd 3 , where m is the number of segments for a molecule and d the hard-sphere diameter
for each segment. According to the equation of Cotterman et al. [19] based on the BH perturbation theory,
the relationship between the hard-sphere diameter (d) and the soft-sphere diameter (σ ) can be expressed
as follows:
d 1 + 0.2977 (kT/ε)
= (14)
σ 1 + 0.33163 (kT/ε) + 0.001047 (kT/ε)2
where ε/k is the Lennard–Jones energy parameter of dispersion energy.
For Lennard–Jones fluid, F LJ is calculated by use of the expression of Cotterman et al. [19]
 
F LJ 1 1 LJ
=m F1 + F2
LJ
(15)
NkT Tr Tr
where
F1LJ = ρr (−8.5859 − 4.5424ρr − 2.1268ρr2 + 10.285ρr3 )
F2LJ = ρr (−1.9075 + 9.9724ρr − 22.216ρr2 + 15.904ρr3 )

where Tr = (kT)/ε and ρr = (6/ 2π)η.
The term of Fchain is calculated as follows [17]:
F chain
= (1 − m) ln [g hs (d)] (16)
NkT
where g hs (d) is the radial distribution function of hard-sphere system, which can be expressed as follows:
1 − 0.5η
g hs (d) =
(1 − η)2
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 759

For dipole–dipole interaction, we use the perturbation equations of Cong et al. [20]. In order to make
the convergence of the calculation, the Pade approximation suggested by Stell et al. [21] is used.
F dd F2dd /(NkT)
= (17)
NkT (1 − F3dd )/F2dd
where
F2dd 2π F3ddd 4π 2 3 2 6
= − 3 β 2 ρµ4 Idd , = β ρ µ Iddd
NkT 3d NkT 27d 3
where β = 1/(kT) and µ is the dipole moment,
1 + 0.18158ρ ∗ − 0.11467ρ ∗2 5(1 + 1.12754ρ ∗ + 0.56192ρ ∗2 )
Idd = , Iddd =
3(1 − 0.49303ρ ∗ + 0.06293ρ ∗2 ) 24(1 − 0.05495ρ ∗ + 0.13332ρ ∗2 )
where ρ ∗ = mρd 3 .
For associating fluids, the Helmholtz energy change Fassoc can be expressed as follows [17]:
 
F assoc  XA 1
= ln X −
A
+ M (18)
NkT A
2 2

where M is the number of association sites on each molecule. The M values are taken from the same as
those
 from Huang and Radosz [17]. The sum over all the associating sites on the molecule is represented
by A and XA is the mole fraction of molecules not bonded at site A, which can be expressed as follows:
 −1

X = 1 + N0 ρX 
A B AB
(19)
B

where N0 is the Avogadro constant and AB stands for the association strength which can be expressed
as follows:
  AB  
ε
 = g(d)
AB seg
exp − 1 (d 3 κ AB ) (20)
kT
where κ AB and εAB /k stand for the volume and the energy of association, respectively. And g(d)seg is
the segmental radial distribution function, which can be approximated as follows:
1 − 0.5η
g(d)seg ≈ g hs (d) =
(1 − η)2
The compressibility factor Z for non-associating polar fluids can be expressed as follows:
Z − Z id = Z hs + Z LJ + Z chain + Z dd (21)
where
 
4η − 2η2 2.5η − η2 Z1LJ Z2LJ
Z = 1,
id
Z =m
hs
, Z chain
= (1 − m) , Z LJ
=m + 2
(1 − η)3 (1 − η)(1 − 0.5η) Tr Tr
Z1LJ = ρr (−8.5859 − 2 × 4.5424ρr − 3 × 2.1268ρr2 + 4 × 10.285ρr3 )
760 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

Z2LJ = ρr (−1.9075 + 2 × 9.9724ρr − 3 × 22.216ρr2 + 4 × 15.904ρr3 )

Z2dd + Z3ddd − 2Z2dd (F3ddd /F2dd )


Z dd =
((1 − F3ddd )/F2dd )2
where
   
F2dd dIdd F3ddd dIddd
Z2dd= 1+ , Z3 =
ddd
2+
NkT Idd NkT Iddd
   
∂Idd ∂Iddd
dIdd = ρ , dIddd = ρ
∂ρ T ,N ∂ρ T ,N

For associating fluids, the compressibility factor Zassoc is used instead of Zdd in Eq. (21) as follows:
 1 
1 ∂XA
Z assoc
=ρ −
A
XA 2 ∂ρ

3. Calculation and discussion

3.1. Correlation of segment parameters

The segment parameters m, σ , ε/k, εAB /k and κ AB for associating fluids and m, σ and ε/k for
non-associating polar fluids can be obtained by simultaneously fitting the experimental saturated vapor
pressures pv and liquid densities pl . The regressed segment parameters with their correlated deviations
for associating and non-associating liquids are shown in Tables 1 and 2, respectively.

Table 1
Regressed segment parameters and the average relative deviations (ARD, %) of pv and pl for pure associating fluidsa
Compound Temperature range (K) Segment parameters ARD (%) of pv ARD (%) of ρ l

m σ (10−10 m) ε/k (K) ε AB /k (K) κ AB M

CH3 OH 337.9–493.2 1.579 3.147 230.0 2320.9 0.062 2 0.52 1.73


C2 H5 OH 351.5–483.0 2.245 3.150 213.5 2391.7 0.035 2 0.54 1.04
C3 H7 OH 373.0–523.2 2.745 3.220 216.5 2413.5 0.023 2 1.24 1.65
C4 H9 OH 334.5–505.0 3.133 3.353 225.4 2513.4 0.011 2 2.75 0.60
C5 H11 OH 283.15–483.15 3.496 3.397 220.5 2622.4 0.012 2 2.03 1.58
C8 H17 OH 283.15–483.15 4.296 3.621 230.7 2813.1 0.009 2 1.77 2.07
C9 H19 OH 283.15–483.15 4.996 3.964 210.9 2865.1 0.016 2 1.11 1.66
C10 H21 OH 283.15–483.15 5.350 4.107 212.3 2937.8 0.013 2 1.53 2.47
H2 O 283.15–583.15 1.055 2.974 483.2 1621.9 0.029 3 1.90 3.04
H2 S 212.15–320.15 1.258 3.393 250.1 508.2 0.010 3 2.03 3.02
NH3 210.15–340.10 1.195 3.081 270.1 798.3 0.031 3 1.16 3.33
Total average deviation (%) 1.5 2.1
a
Experimental data are taken from Beaton and Hewitt [22].
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 761

Table 2
Regressed segment parameters and the average relative deviations (ARD, %) of pv and ρ l for pure non-associating polar fluidsa
Compound Temperature Dipole moment, Segment parameters ARD (%) ARD (%)
range (K) ␮ (10−30 C N)b of pv of ρ l
m σ (10−10 m) ε/k (K)

CHCl3 334.5–505.0 3.47 2.751 3.310 225.4 1.09 1.21


CHClF2 242.4–355.0 4.74 2.664 3.142 203.7 1.62 1.88
CClF3 191.7–295.0 1.67 2.603 3.129 133.5 1.04 1.37
CO 81.66–125.0 0.37 1.000 3.718 110.8 4.57 3.24
Total average deviation (%) 2.1 1.9
a
Experimental data are taken from Beaton and Hewitt [22].
b
Dipole moment data are taken from [23].

3.2. Calculation of density profile ρ(z) and surface thickness t

The fluid in the surface region is considered as an non-homogeneous one. In the horizontal directions
in the surface, it is assumed no density gradient exists. However, in the vertical direction of the surface,
there exists the density gradient, that is, dρ(z)/dz = 0. According to the molecules simulation data, the
function between ρ(z) and z is given as a hyperbolic tangent function [24,25].
 
1 l 1 l (z − z0 )
ρ(z) = (ρ + ρ ) − (ρ + ρ ) tanh 2
v v
(22)
2 2 t

where ρ l and ρ v are the liquid and vapor densities, respectively, t stands for the surface thickness and z0
is given by,

ρ(z0 ) = 21 (ρ l + ρ v )

By using the parameters in Tables 1 and 2, the liquid densityρ l and vapor density ρ v at a specific
temperature in equilibrium can be calculated by the following equations:

(F + pV)l = (F + pV)v
(23)
p = Z l ρ l kT = Z v ρ v kT
The surface region is divided into many layers, the density in any layer is treated as the average density.
zi+1
z ρ(z) dz
ρ(zi ) = i (i = 1, · · · , n) (24)
zi+1 − zi
where the subscript i is the number of the layer and n the total number of the layers.
The surface thickness t, plays an important role in the calculation of density profile and the surface
tension. In this paper, it is calculated by the equilibrium principle that the chemical potentials in every
layer of the surface must be equal to each other.

µ(1) = µ(2) = · · · = µ(i) = · · · = µ(n)


762 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

Fig. 2. Density profiles of methanol from 337.2 to 493.2 K.

The objective function used to optimize the surface thickness t is as follows:


n
 (µ(i) − µbulk )2
fs = (25)
i=1
n

where µbulk is the chemical potential of the two bulk phases, which is obtained simultaneously with the
liquid and vapor densities by solving Eq. (23). The chemical potentials in every layer µ(i) (1 ≤ i ≤ n)
can be calculated by use of the regressed parameters and the calculated average density ρ(zi ).
By combining Eqs. (22), (24) and (25), the density profile and the surface thickness can be calculated.
We found the surface should be divided into at least 2000 layers, in this case the chemical potentials in
every layer change no longer and equal to each other. Thus, the calculated density profile and the surface
thickness can be considered as the equilibrium density profile and equilibrium thickness, respectively.
By use of this method, the density profile and the surface thickness are obtained simultaneously.
The density profiles of methanol from 337.2 to 493.2 K are shown in Fig. 2. This figure shows that both
the density difference between liquid and vapor phases and the slope of density profile change smaller
when the temperature increases. The surface thickness of methanol from 337.2 to 493.2 K is shown in
Fig. 3. From this figure and Table 1 it can be seen that the calculated surface thickness is several times of
the hard-sphere diameter and increases with temperature.

3.3. Calculation of surface tension γ

By using the obtained density ρ0 (zi ), the surface tension can be calculated as follows:
 ∞ 
2000
γ = [p − p[ρ0 (z)]] dz ≈ [p − p[ρ0 (zi )]] zi (26)
−∞ i=1
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 763

Fig. 3. Surface thickness of methanol from 337.2 to 473.2 K.

where zi is the thickness of every layer in the surface, zi = 0.005 × 10−10 m. The normal pressure
tensor,ρ0 (zi ) is calculated from Eq. (22) and tangential pressure tensor p[ρ0 (zi )] is calculated from Eq. (9)
by use of ρ0 (zi ) and the parameters in EOS. The difference between the normal and tangential pressure
tensors is given by p −p[ρ0 (zi )], the integral of which forms the surface tension. The values p −p[ρ0 (zi )]
of methanol from 337.2 to 473.2 K is shown in Fig. 4. This figure shows that value of p − p[ρ0 (zi )] has

Fig. 4. The difference between normal and tangential pressure tensors of methanol from 337.2 to 473.2 K.
764 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

Table 3
Prediction of the surface tension for pure fluids
Compound Data points of γ ARD (%) of γ Data sources

CH3 OH 8 3.77 [22]


C2 H5 OH 8 2.52 [22]
C3 H7 OH 9 4.27 [22]
C4 H9 OH 9 3.06 [22]
C5 H11 OH 10 2.59 [27]
C8 H17 OH 7 2.14 [27]
C9 H19 OH 10 2.78 [27]
C10 H21 OH 10 2.66 [27]
H2 O 27 4.46 [22]
H2 S 7 2.32 [22]
NH3 7 2.64 [22]
CO 8 2.51 [22]
CHCl3 9 2.90 [22]
CHClF2 8 2.44 [22]
CClF3 8 1.98 [22]
Total average deviation (%) 2.9

a maximum in the surface region and its absolute value decreases when the temperature increases, which
has been proved by the molecular dynamics simulation [26].
By using Eq. (26), the surface tensions of 15 pure polar liquids are predicted. The predicted surface
tensions agree very well with the experimental data, which have been shown in Table 3 and Figs. 5–10.
The total average relative deviation of prediction is 2.9%. These results show that our results coincide
with the experimental data.

Fig. 5. Predicted and experimental surface tension for methanol and 1-propanol ((䉬, 䊉) experimental data, (—) predicted data).
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 765

Fig. 6. Predicted and experimental surface tension for ethanol and 1-butanol ((䉬, 䊉) experimental data, (—) predicted data).

Fig. 7. Predicted and experimental surface tension for pentanol, octanol, nonanol and decanol ((䉬, 䊉) experimental data, (—)
predicted data).
766 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

Fig. 8. Predicted and experimental surface tension for water, ammonia and hydrogen sulfide ((䉬, 䊉) experimental data, (—)
predicted data).

Fig. 9. Predicted and experimental surface tension for CO, and CClF3 ((䉬, 䊉) experimental data, (—) predicted data).
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 767

Fig. 10. Predicted and experimental surface tension for CClF2 , and CHCl3 ((䉬, 䊉) experimental data, (—) predicted data).

4. Conclusion

1. The BH perturbation theory and SAFT are applied to build the EOS for pure polar and associating
fluids. The segment parameters m, ε/k and σ for pure non-associating polar fluids and m, σ , ε/k, ε AB /k
and κ AB for associating fluids are obtained by simultaneously fitting the experimental saturated vapor
pressures and liquid densities. The correlation deviations of the vapor pressure and liquid density are
2.1 and 1.9% for four non-associating polar fluids, 1.5 and 2.1% for 11 associating fluids, respectively.
2. The DFT is applied in this work. By dividing the surface region into 2000 layers, the number density
ρ0 (zi ) and the pressure p[ρ0 (zi )] in every layer of the surface are calculated. The surface thickness for
15 pure polar and associating fluids is calculated, and the results show it increases with the temperature.
3. The surface tensions of 15 pure polar and associating fluids are predicted with the same set of parameters
in EOS. The average deviation of prediction is 2.9%.
4. Our method has good correlation capabilities for pVT, density profile and surface thickness and also
has good prediction for surface tension of pure polar fluids and associating fluids.

List of symbols
A surface area (nm2 )
d hard-sphere diameter (nm)
Eext external potential (J)
F Helmholtz free energy (J)
f Helmholtz free energy density (J m−3 )
g radial distribution function
k Boltzmann constant (J K−1 )
M the number of association sites on each associating molecule
768 J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769

m number of segments for one molecule


N number of molecules
n total number of the layers
N0 Avogadro’s constant (mol−1 )
p pressure (Pa)
drr infinitesimal (m3 )
T absolute temperature (K)
V volume (m3 )
XA mole fraction of molecules not bonded at site A
Z compressibility factor
z distance in vertical direction (nm)

Greek letters
γ surface tension (mN m−1 )
 association strength
ε dispersion energy parameter (J)
η packing factor
κ association volume
µ chemical potential (J mol−1 ); dipole moment (C N)
ρ number density (m−3 )
ρ0 equilibrium number density (m−3 )
σ soft-sphere diameter (nm)
Ω grand potential (J)

Superscripts
assoc associating
chain hard-sphere chain
dd dipole–dipole
hs hard sphere
id ideal
LJ Lennard–Jones
l liquid
v vapor

Subscripts
exp experimental values
ext external
i number of layers
r relative value
J.-F. Lu et al. / Fluid Phase Equilibria 194–197 (2002) 755–769 769

Acknowledgements

The financial supports of this work by the National Natural Science Foundation of China (no. 29736170),
the Fundamental Research Fund of Tsinghua University in China (no. JC1999038) and the Scientific
Research Foundation for Returned Overseas Chinese Scholars, Ministry of Education, China are gratefully
appreciated.

References

[1] F.B. Sprow, J.M. Prausnitz, Can. J. Chem. Eng. 45 (1967) 25–28.
[2] Z.-B. Li, Y.-G. Li, J.-F. Lu, Ind. Eng. Chem. Res. 38 (1999) 1133–1139.
[3] A.I. Rusanov, Pure Appl. Chem. 64 (1992) 111–124.
[4] S. Toxvaerd, J. Chem. Phys. 55 (1971) 3116–3120.
[5] J.M. Haile, C.G. Gray, K.E. Gubbins, J. Chem. Phys. 64 (1976) 2569–2578.
[6] D. Fu, J.-F. Lu, T.-Z. Bao, Y.-G. Li, Ind. Eng. Chem. Res. 39 (2000) 320–327.
[7] R. Evans, Adv. Phys. 28 (1979) 143–200.
[8] J. Winkelmann, U. Brodrecht, I. Kreft, Ber. Bunsenges. Phys. Chem. 98 (1994) 912–919.
[9] J. Winkelmann, Ber. Bunsenges. Phys. Chem. 98 (1994) 1308–1316.
[10] T. Wadewitz, J. Winkelmann, Ber. Bunsenges. Phys. Chem. 100 (1996) 1825–1832.
[11] P.I. Teixeira, M.M. Telo da Gama, J. Phys. Cond. Matt. 3 (1991) 111–125.
[12] P. Tarazona, U.M.B. Marconi, R. Evans, Mol. Phys. 60 (1987) 573–595.
[13] B.K. Peterson, K.E. Gubbins, G.S. Heffelfinger, U.M.B. Marconi, F.J. van Swol, Chem. Phys. 88 (1988) 6487–6500.
[14] J. Cai, H.-L. Liu, Y. Hu, J. East China Univ. Sci. Tech. 26 (2000) 100–102 (in Chinese).
[15] J.A. Barker, D. Henderson, J. Chem. Phys. 47 (1967) 4714–4721.
[16] W.G. Chapman, K.E. Gubbins, G. Jackson, M. Radosz, Ind. Eng. Chem. Res. 29 (1990) 1709–1723.
[17] S.H. Huang, M. Radosz, Ind. Eng. Chem. Res. 29 (1990) 2284–2296.
[18] N.F. Carnahan, K.E. Starling, J. Chem. Phys. 51 (1969) 635–636.
[19] R.L. Cotterman, B.J. Schwarz, J.M. Prausnitz, AIChE J. 32 (1986) 1787–1798.
[20] W. Cong, Y.-G. Li, J.-F. Lu, Fluid Phase Equilibria 124 (1996) 55–65.
[21] G. Stell, J.C. Rasaiah, H. Narang, Mol. Phys. 27 (1974) 1393–1414.
[22] C.F. Beaton, G.F. Hewitt, Physical Property Data for the Design Engineer, Hemisphere, New York, 1989.
[23] K.H. Hellwege, in: L. Bornstein (Ed.), Numerical Data and Functional Relationships in Science and Technology, Group II,
Molecular Constants, Vol. 6, Springer, Heidelberg, 1974.
[24] J.S. Rowlinson, B. Widom, Molecular Theory of Capillarity, Clarendon Press, Oxford, 1982.
[25] M.M. Telo da Gama, R. Evans, Mol. Phys. 38 (1979) 367–375.
[26] A. Jose, D.J. Tildesley, G.A. Chapela, J. Chem. Phys. 102 (1995) 4574–4583.
[27] J.J. Jasper, J. Phys. Chem. Ref. Data 1 (1972) 841–1009.

You might also like